We describe a fully automated processing pipeline to support the noninvasive absolute quantification of the cerebral metabolic rate for glucose (CMRGlc) in a clinical setting. This pipeline takes advantage of “anatometabolic” information associated with fully integrated PET/MRI. Methods: Ten healthy volunteers (5 men and /5 women; 27 ± 7 y old; 70 ± 10 kg) underwent a test-retest ¹⁸F-FDG PET/MRI examination of the brain. The imaging protocol consisted of a 60-min PET list-mode acquisition with parallel MRI acquisitions, including 3-dimensional time-of-flight MR angiography, MRI navigators, and a T1-weighted MRI scan. State-of-the-art MRI-based attenuation correction was derived from T1-weighted MRI (pseudo-CT [pCT]). For validation purposes, a low-dose CT scan was also performed. Arterial blood samples were collected as the reference standard (arterial input function [AIF]). The developed pipeline allows the derivation of an image-derived input function (IDIF), which is subsequently used to create CMRGlc maps by means of a Patlak analysis. The pipeline also includes motion correction using the MRI navigator sequence as well as a novel partial-volume correction that accounts for background heterogeneity. Finally, CMRGlc maps are used to generate a normative database to facilitate the detection of metabolic abnormalities in future patient scans. To assess the performance of the developed pipeline, IDIFs extracted by both CT-based attenuation correction (CT-IDIF) and MRI-based attenuation correction (pCT-IDIF) were compared with the reference standard (AIF) using the absolute percentage difference between the areas under the curves as well as the absolute percentage difference in regional CMRGlc values. Results: The absolute percentage differences between the areas under the curves for CT-IDIF and pCT-IDIF were determined to be 1.4% ± 1.0% and 3.4% ± 2.6%, respectively. The absolute percentage difference in regional CMRGlc values based on CT-IDIF and pCT-IDIF differed by less than 6% from the reference values obtained from the AIF. Conclusion: By taking advantage of the capabilities of fully integrated PET/MRI, we developed a fully automated computational pipeline that allows the noninvasive determination of regional CMRGlc values in a clinical setting. This methodology might facilitate the proliferation of fully quantitative imaging into the clinical arena and, as a result, might contribute to improved diagnostic efficacy.

我们描述了一个全自动的处理管道，以支持在临床环境中葡萄糖（CMRGlc）的脑代谢率的无创绝对定量。该管道利用了与完全集成的PET / MRI相关的“代谢”信息。方法：十名健康志愿者（5名男性和/ 5名女性； 27±7岁； 70±10 kg）进行了重新测试¹⁸大脑的F-FDG PET / MRI检查。成像协议包括60分钟PET列表模式采集和并行MRI采集，包括3维飞行时间MR血管造影，MRI导航仪和T1加权MRI扫描。基于T1加权MRI（伪CT [pCT]）得出了基于MRI的最新衰减校正。为了验证，还进行了低剂量CT扫描。收集动脉血样本作为参考标准（动脉输入功能[AIF]）。开发的管道允许派生图像输入函数（IDIF），随后将其用于通过Patlak分析创建CMRGlc映射。该管道还包括使用MRI导航器序列进行的运动校正以及考虑到背景异质性的新颖的部分体积校正。最后，CMRGlc映射用于生成规范数据库，以利于在将来的患者扫描中检测代谢异常。为了评估已开发管道的性能，将通过基于CT的衰减校正（CT-IDIF）和基于MRI的衰减校正（pCT-IDIF）提取的IDIF与参考标准（AIF）进行了比较，使用曲线下的面积以及区域CMRGlc值的绝对百分比差。结果： CT-IDIF和pCT-IDIF曲线下面积之间的绝对百分比差异分别确定为1.4％±1.0％和3.4％±2.6％。基于CT-IDIF和pCT-IDIF的区域CMRGlc值的绝对百分比差异与从AIF获得的参考值的差异小于6％。结论：通过利用完全集成的PET / MRI的功能，我们开发了一种全自动的计算管线，可以在临床环境中无创确定区域CMRGlc值。这种方法可能会促进完全定量成像进入临床领域的扩散，并因此可能有助于提高诊断效力。